StartsWith¶

Overview¶

The StartsWith expression is a string predicate that determines whether a string value begins with a specified prefix. It extends the StringPredicate class and supports collation-aware string comparison for case-sensitive and case-insensitive matching.

Syntax¶

string_expr LIKE 'prefix%'
STARTSWITH(string_expr, prefix)

// DataFrame API
col("column_name").startsWith("prefix")

Arguments¶

Argument	Type	Description
left	Expression	The string expression to check
right	Expression	The prefix string expression to match against

Return Type¶

BooleanType - Returns true if the left string starts with the right string, false otherwise.

Supported Data Types¶

StringType with non-CSAI (Case-Sensitive Accent-Insensitive) collations
Supports trim collation for string normalization
Both arguments must be string-compatible types

Algorithm¶

Delegates string comparison to CollationSupport.StartsWith.exec() method
Uses the expression's collationId for collation-aware comparison
Performs UTF8String-based comparison for efficient string operations
Handles collation rules including case sensitivity and accent handling
Returns boolean result based on prefix matching logic

Partitioning Behavior¶

This expression does not affect partitioning behavior:

Preserves existing partitioning schemes
Does not require data shuffle operations
Can be applied as a filter predicate without repartitioning

Edge Cases¶

Null handling: If either argument is null, the result is null
Empty string behavior: Empty prefix ("") returns true for any string
Empty left string with non-empty prefix returns false
Collation-specific edge cases depend on the configured collation rules
Case sensitivity depends on the collation configuration

Code Generation¶

This expression supports Tungsten code generation through the doGenCode method:

Uses CollationSupport.StartsWith.genCode() for optimized code generation
Falls back to interpreted mode if code generation fails
Generates efficient bytecode for runtime string comparison

Examples¶

-- Check if customer names start with 'John'
SELECT * FROM customers WHERE name LIKE 'John%';

-- Using STARTSWITH function (if available)
SELECT STARTSWITH(product_name, 'iPhone') FROM products;

// DataFrame API usage
import org.apache.spark.sql.functions._

// Filter rows where column starts with prefix
df.filter(col("name").startsWith("John"))

// Select with startsWith condition
df.select(col("*")).where(col("product_name").startsWith("iPhone"))